Subscriber access provided by University of Sydney Library
Article
Field study of rice yield diminished by soil arsenic in Bangladesh Brittany Lynn Huhmann, Charles F Harvey, Anjal Uddin, Imtiaz Choudhury, Kazi M. Ahmed, John M Duxbury, Benjamin C. Bostick, and Alexander Van Geen Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.7b01487 • Publication Date (Web): 20 Sep 2017 Downloaded from http://pubs.acs.org on September 21, 2017
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 27
1
Environmental Science & Technology
Field study of rice yield diminished by soil arsenic in Bangladesh
2 3
Brittany L. Huhmann,*,1 Charles F. Harvey,1 Anjal Uddin,2 Imtiaz Choudhury,2 Kazi M.
4
Ahmed,2 John M. Duxbury,3 Benjamin C. Bostick,4 and Alexander van Geen4
5 6
1
7
Cambridge, MA, 02139, USA
8
2
Department of Geology, University of Dhaka, Dhaka, 1000, Bangladesh
9
3
School of Integrative Plant Science, Cornell University, Ithaca, NY, 14850, USA
10
4
Lamont-Doherty Earth Observatory, Columbia University, Palisades, NY, 10964, USA
Department of Civil and Environmental Engineering, Massachusetts Institute of Technology,
11 12
TOC art:
13 14 15 16
Abstract Rice was traditionally grown only during the summer (aman) monsoon in Bangladesh but
17
more than half is now grown during the dry winter (boro) season and requires irrigation. A
18
previous field study conducted in a small area irrigated by a single high-arsenic well has shown
19
that the accumulation of arsenic (As) in soil from irrigating with high-As groundwater can
ACS Paragon Plus Environment
Environmental Science & Technology
20
reduce rice yield. We investigated the effect of soil As on rice yield under a range of field
21
conditions by exchanging the top 15 cm of soil between 13 high-As and 13 low-As plots
22
managed by 16 different farmers, and we explore the implications for mitigation. Soil As and
23
rice yields were measured for soil replacement plots where the soil was exchanged and adjacent
24
control plots where the soil was not exchanged. Differences in yield (ranging from +2 to -2 t/ha)
25
were negatively correlated to the differences in soil As (ranging from -9 to +19 mg/kg) between
26
adjacent replacement and control plots during two boro seasons. The relationship between soil
27
As and yield suggests a boro rice yield loss over the entire country of 1.4-4.9 million tons
28
annually, or 7-26% of the annual boro harvest, due to the accumulation of As in soil over the past
29
25 years.
30 31 32
Introduction Much of the groundwater in Bangladesh is contaminated with high levels of arsenic (As)
33
that harm human health when this water is used for drinking and, to a lesser extent, when
34
groundwater is used for irrigating rice and As is taken up by the rice grain1–4. Winter season
35
(boro) rice, grown in standing water maintained by groundwater irrigation, is the dominant crop
36
in Bangladesh in both production5 and caloric consumption6. The groundwater used for irrigation
37
often contains high concentrations of As that accumulates in soil over time and can be taken up
38
by rice plants, elevating As concentrations in rice straw, husk, and grain7–13. Monsoon season
39
(aman) rice is often grown in the same fields where boro rice is cultivated, and although it is
40
primarily rainfed during the monsoon, it is still exposed to the high concentrations of soil As that
41
build up during boro irrigation14.
ACS Paragon Plus Environment
Page 2 of 27
Page 3 of 27
42
Environmental Science & Technology
Elevated As concentrations in irrigation water and soil have been found to decrease boro
43
and aman rice yield in greenhouse studies and pot experiments8,9,15–20. A prior field study in
44
Faridpur, Bangladesh found that boro rice yields were 7-9 t/ha where soil As concentrations were
45
low (~10 mg/kg), but were much poorer, 2-3 t/ha, where soil As concentrations were high (~70
46
mg/kg)21.
47
While greenhouse and pot studies have shown the negative effects of As on rice yield,
48
these studies do not provide sufficient information to quantify the magnitude of the yield impact
49
of As under field conditions. Furthermore, the only previous field study on the yield effects of As
50
did not include aman-season rice and was conducted in an 8 ha area managed by a single farmer
51
and irrigated by a single high-As well – and thus under a relatively narrow set of conditions. The
52
goal of our study was to quantify the yield impacts of As under a broader array of field
53
conditions by using a controlled study design. To quantify the yield impact of As we exchanged
54
high- and low-As soils at thirteen field sites distributed throughout a 150 km2 area in Faridpur
55
district, Bangladesh and compared these soil replacement plots to adjacent control plots. Our
56
study plots were managed by sixteen different farmers, and these farmers chose to cultivate two
57
boro rice varieties and nine aman rice varieties. We hypothesized that replacing high-As soil
58
with low-As soil would improve yield, and that replacing low-As soil with high-As soil would
59
cause a decline in yield. We tested this hypothesis for rice grown during the 2015 and 2016 boro
60
and aman seasons.
61 62
Materials and Methods
63
Experimental Site and Design
ACS Paragon Plus Environment
Environmental Science & Technology
64
The study was conducted in fields irrigated by high-As wells in Faridpur district,
65
Bangladesh (Figure 1). The wells drew water from 40 to 100 m in depth, ranged from 5 to 43
66
years in age, and had As concentrations of 100 to 400 µg/L as measured by the ITS Econo-Quick
67
field kit and converted based on a prior intercalibration of field kit As concentrations with As
68
concentrations measured by ICP-MS22 (Table S1). As has been observed at other field sites, soil
69
As tended to decrease away from the irrigation inlet, which is where much of the reduced iron in
70
the irrigation water precipitates to form iron oxides that adsorb or coprecipitate As12,21,23.
71
Up to two rice crops are grown at our study sites each year. Boro rice is grown during the
72
dry season, from mid-January through May, and is irrigated with groundwater. During the 2015
73
and 2016 boro seasons, farmers at our study sites grew two rice varieties, BRRI dhan 28 (BR 28)
74
and BRRI dhan 29 (BR 29). These are also the predominant rice varieties grown across
75
Bangladesh, and were estimated in 2005 to be grown in nearly 60% of the total boro rice cropped
76
area in the country24. Aman rice is traditionally grown during the monsoon season, from June
77
through mid-November, and is primarily rainfed, with supplemental groundwater irrigation if
78
needed. While boro rice is always transplanted, aman rice may be transplanted or broadcast
79
sown. During the aman season, farmers grow a larger number of rice varieties. At our study sites,
80
farmers grew nine different varieties during the aman 2015 season and five different varieties
81
during the aman 2016 season. Many of these were local varieties, but the dominant variety
82
grown during both aman seasons was BR 39.
83
In January 2015 before the fields were transplanted with boro rice, we exchanged the top
84
fifteen centimeters of soil between thirteen 5×5 m high-As (near the irrigation inlet) and low-As
85
(far from the irrigation inlet) plots (Figure 1). Each plot where soil was replaced was paired with
86
an adjacent control plot where the soil remained undisturbed and no changes were made. Soil
ACS Paragon Plus Environment
Page 4 of 27
Page 5 of 27
Environmental Science & Technology
87
was swapped within a field in four cases, and swapped between nearby fields in nine cases
88
(Figure S1). By pairing each 5×5 m soil exchange plots with an adjacent 5×5 control plot, we
89
implemented a study design that controls for the management of the plots (the plots in each pair
90
are managed by the same farmer, fertilized and irrigated in the same way, and planted with the
91
same rice variety) and for the environmental conditions (air temperature, relative humidity,
92
sunlight, rainfall). We then measured soil As concentrations and rice yields in the soil
93
replacement and control plots during the 2015 and 2016 boro and aman rice seasons.
94 95 96
Soil As Measurements Total soil As concentrations were measured using an Innov-X Delta Premium field X-ray
97
fluorescence (XRF) spectrometer in the soil mode for a total counting time of 35-150 s. Soil
98
standards 2709 and 2711 from the National Institute of Standards and Technology (NIST) were
99
analyzed at the beginning and end of each day and periodically during longer sample runs. The
100
measured average and standard deviation for standard 2711 of 103 ± 7 (n = 39) matched the
101
reference value of 105 ± 8 mg/kg. The measured average and standard deviation for standard
102
2709 of 16.4 ± 1.9 (n = 19) matched the reference value of 17.7 ± 0.8 mg/kg. All soil As
103
concentrations were above the detection limit of the XRF.
104
At harvest time, twelve 20-cm deep soil cores (diameter of 3 cm) were collected from
105
each plot where rice yield was measured. Three soil cores were collected at a distance of 1 m
106
inward from each of the four sides of a 5 × 5 m plot and combined, for a total of four composited
107
soil samples from each plot. The average and standard error of these four samples were used to
108
represent the soil arsenic concentration in each plot. The soil samples were dried in an oven at
ACS Paragon Plus Environment
Environmental Science & Technology
109
40ºC and homogenized by mortar and pestle before As analysis with XRF, to ensure a moisture
110
content and sample morphology similar to the NIST standards used 25.
111
Single 20-cm deep soil cores (diameter of 3 cm) were also collected monthly from each
112
plot from anywhere in the 5×5 m area throughout each growing season. These soil samples were
113
dried in an oven at 40ºC and homogenized by mortar and pestle in 5 cm increments from 0 to 20
114
cm, and those from the boro 2016 and aman 2016 growing seasons were analyzed by XRF to
115
provide depth profiles of soil As, two examples of which are shown in Figure 2.
116 117 118
Measurements of Additional Soil Element Concentrations Concentrations of K, Ca, Ti, Cr, Mn, Fe, Ni, Cu, Zn, Rb, Sr, and Zr were also measured
119
by XRF during the course of the As measurements. The measured averages for the NIST 2709
120
standard were within 12% of the reported values (Table S3).
121 122 123
Soil Nutrient Measurements Samples from the single 20 cm cores collected monthly from each plot during the boro
124
2015 season were also sent to the BRAC soil laboratory in Gazipur, Bangladesh, for
125
measurement of electrical conductivity (measured on a 1:1 mixture of soil and distilled water), N
126
(total Kjeldahl nitrogen), organic carbon (Walkley-Black method), P (modified Olsen method),
127
K (ammonium acetate extraction), S (calcium hydrogen phosphate extraction), and Zn
128
(diethylenetriaminepentaacetic acid extraction). Nutrients were measured similarly during the
129
subsequent boro and aman seasons, except that sets of three 20 cm cores (rather than a single
130
core) were collected monthly from each plot to ensure sufficient soil for nutrient analyses.
131
ACS Paragon Plus Environment
Page 6 of 27
Page 7 of 27
132 133
Environmental Science & Technology
Rice Yield Measurements Rice yields were measured for a 3×3 m area in the center of each 5×5 m plot. The rice
134
was threshed immediately after harvest, its weight and moisture content were recorded, and yield
135
values were adjusted to 14% moisture content. During the boro 2016 season, we obtained an
136
estimate of the error on yield by dividing each 3×3 m plot along the diagonal and making a
137
separate measurement of the yield for each half of the 3×3 plot.
138
Due to miscommunication with the farmers and other farmer decisions, yield
139
measurements were obtained for only a subset of the 26 plots at the end of each growing season.
140
We obtained yield measurements for 13 pairs of soil replacement and control plots during the
141
boro 2015 season, 24 pairs during the aman 2015 season, 17 pairs during the boro 2016 season,
142
and 19 pairs during the aman 2016 season (Table S4).
143 144
Results
145
Effect of the Soil Exchange on Soil As and Soil Nutrients
146
The soil exchanges had a large effect on soil As concentrations at some study sites
147
(Figure 2a) and a minimal effect at others (Figure 2b). In some of the cases where the effect was
148
small, the soil exchange was conducted after some initial irrigation and because the soil was very
149
wet it was hard to ensure that precisely the top 15 cm of soil were exchanged. For the boro 2015
150
growing season, we observed that the soil replacement on average decreased soil As for the high-
151
As plots (-4.0 ± 3.5 mg/kg) and increased soil As for the low-As plots (+12 ± 3 mg/kg) compared
152
to the adjacent control plots (Figure 3a). This represents an average 8% decrease in soil As
153
concentration for the high-replaced-by-low plots and a 65% increase in As concentration for the
154
low-replaced-by-high plots compared to their respective control plots. We did not observe a
ACS Paragon Plus Environment
Environmental Science & Technology
155
significant difference between replaced and control plots for OC, N, P, K, S, Zn, or EC during
156
the boro 2015 growing season or thereafter (Figure S2, Table S5).
157
The effect of the soil replacement on soil As remained significant for plots observed
158
during the aman 2015 growing season, with a -4 ± 3 mg/kg (11%) decrease observed in the high-
159
As plots and a 4 ± 1 mg/kg (20%) increase observed in the low-As plots. However, the
160
difference in soil As between high-As and low-As pairs of plots was no longer detectable during
161
the boro and aman 2016 seasons (Figure 3a).
162 163
Effect of the Soil Exchange on Rice Yield
164
For the boro 2015 growing season, the soil replacement increased yield for the high-As
165
plots (+0.8 ± 0.4 t/ha) and decreased yield for the low-As plots (-0.47 ± 0.45 t/ha) compared to
166
the adjacent control plots (Figure 3b). This represents a 16% increase in yield for the high-
167
replaced-by-low plots and a 6.6% decrease in yield for the low-replaced-by-high plots as
168
compared with their respective control plots. The average yield differences between pairs of
169
high-As soil replacement and control plots and pairs of low-As soil replacement and control plots
170
significantly differed from each other at the p = 0.05 level.
171
Unlike with soil As, the effect of the soil replacement on yield remained significant for
172
plots observed during both the aman 2015 and boro 2016 seasons, with the high-replaced-by-low
173
plots continuing to show an increase in yield and the low-replaced-by-high plots a decrease in
174
yield compared to their control plots. During the aman 2015 season, the replacement of high-As
175
soil with low-As soil achieved on average a yield increase of 8% (yield difference of 0.2 ± 0.1
176
t/ha and average high-As control plot rice yield of 2.59 t/ha). During the boro 2016 season, the
177
replacement of high-As soil with low-As soil achieved a yield increase of 17% (yield difference
ACS Paragon Plus Environment
Page 8 of 27
Page 9 of 27
Environmental Science & Technology
178
of 0.5 ± 0.3 t/ha and average high-As control plot yield of 3 t/ha), which is similar to the 16%
179
increase observed in boro 2015. We did not observe a significant difference in yield between
180
high-As and low-As pairs of plots during the aman 2016 season (Figure 3b).
181 182 183
Yield as a function of soil As There is no direct correlation between yield and soil As for our study plots (Figure S3);
184
other factors that affect yield evidently conceal the effect of soil As. Our experimental design,
185
where each soil replacement plot is paired with an adjacent unaltered control plot, allows us to
186
control for many of these other factors. Computing the difference in soil As and the difference in
187
rice yield between each soil replacement and adjacent control plot holds constant fertilizer and
188
pesticide use, farmer care (e.g. weeding), transplanting and harvesting dates, irrigation water
189
source, rice variety, and other variations in local conditions (e.g. air temperature, relative
190
humidity, sunlight, rainfall).
191
When difference in yield is plotted as a function of difference in soil As between adjacent
192
plots for the boro 2015 season, there is a negative linear relationship with a slope of -0.06
193
(t/ha)(mg/kg)-1 in which soil As accounts for 45% of the variance in yield (Figure 4a). A similar
194
relationship exists between boro 2016 yield difference and boro 2015 soil As difference (R2 =
195
0.87, slope = -0.10) (Figure 4c). During the 2015 boro season the difference in rice yield ranged
196
from +2.3 to -1.9 t/ha (average yield of 6.3 t/ha), and during the boro 2016 season the difference
197
in rice yield ranged from +1.8 to -1.4 t/ha (average yield of 3.8 t/ha). This indicates that
198
exchanging soil caused an increase or decrease by about a third of the average rice yield.
199 200
In contrast, there is no correlation of yield differences observed in aman 2015 or aman 2016 with soil As differences observed in boro 2015 (Figure 4d). Additionally, when using soil
ACS Paragon Plus Environment
Environmental Science & Technology
201
As difference from any season after boro 2015, there is little-to-no correlation with difference in
202
rice yield for any season (Figure S4). As shown by the four pairs of plots where soil As was
203
measured in all four growing seasons (red symbols in Figure S4), this is at least in part because
204
of the decline in the effect of the soil exchange on soil As after the boro 2015 season.
205 206
Yield difference as a multivariable function of As and nutrient concentrations
207
We used a stepwise linear model with a threshold p-value of 0.05 to test whether a
208
number of other factors were significant when added to the regression. These included the
209
differences in OC, N, P, K, S, Zn, and EC between the replaced and control plots as measured in
210
the BRAC lab. These also included differences in total K, Ca, Ti, Cr, Mn, Fe, Ni, Cu, Zn, Rb, Sr,
211
and Zr concentrations as measured by XRF. We also tested the significance of rice variety and,
212
during the aman seasons, whether rice was transplanted or broadcast sown. Finally, we included
213
whether the soil in the replacement plot was swapped from a plot in the same field or a plot in a
214
different field as a binary variable (Figure S5). When we tested these factors across all four
215
seasons, occasionally a factor was significant, but none of the factors were statistically
216
significant during more than one season, and thus none of these factors had a reproducible effect
217
on the observed yield differences between the soil replacement plots and the adjacent control
218
plots. Furthermore, including the other significant factors did not change whether the difference
219
in As was significantly related to difference in yield between pairs of plots. None of these
220
additional measured variables were therefore confounding factors that could explain the
221
observed correlation between soil As difference and yield difference.
222 223
Discussion
ACS Paragon Plus Environment
Page 10 of 27
Page 11 of 27
224 225
Environmental Science & Technology
Effect of soil replacement on soil As Replacing the soil in high-As plots with low-As soil decreased the soil As concentrations,
226
and replacing the soil in low-As plots with high-As soil increased the soil As concentrations,
227
consistent with expectations (Figure 3). Some of the exchanges were more successful than
228
others, however (Figure 2), resulting in an overall modest effect of the exchange on soil arsenic
229
concentrations.
230
Additionally, the effect of the soil exchange on soil arsenic declined over time. We
231
initially hypothesized that this was due to lateral mixing between the soil replacement plots and
232
their surroundings over the course of the two-year experimental study. However, the arsenic
233
content was identical in soil cores taken 1 m from the outer edge and 1 m from the center of the
234
soil replacement plots at the end of the 2017 boro season (Figure S6). Thus, the replacement
235
plots lacked the soil arsenic gradient from edge to center that would be expected if mixing
236
between the study plots and the surrounding soil had occurred. Other possible explanations for
237
the decline in the effect of the exchange on soil arsenic concentrations over time include vertical
238
mixing (since only the top 15 cm of soil were exchanged) or, for the high-arsenic plots, rapid
239
buildup of arsenic in the soil near the irrigation inlet during boro 2015 and boro 2016 irrigation.
240 241
Relationship between soil As difference and yield difference
242
Yield difference between the replacement and control plots in boro 2015 and boro 2016
243
had a strong negative linear dependence on soil As difference as measured in boro 2015 (Figure
244
4). The two rice varieties grown in our study plots during these seasons, BR 28 and BR 29, did
245
not statistically differ in their response to soil As. Panuallah et al. reported slopes of -0.09 and -
ACS Paragon Plus Environment
Environmental Science & Technology
246
0.11 t/ha in 2006 and 2007 for BR 29 respectively, which are comparable but slightly higher than
247
the slopes of -0.06 and -0.10 that we observed in 2015 and 2016.
248
In contrast with boro rice, aman rice yield differences did not correlate strongly with soil
249
As differences (Figure 4). A possible explanation is the larger number of rice varieties that were
250
grown in our study plots during the aman seasons as compared with the boro seasons. Different
251
rice varieties may have different responses to As, and with only a few measurements for each
252
variety, we do not have sufficient information to determine these distinct relationships.
253
Additionally, aman yields are generally lower than boro yields, and our study plots had average
254
aman yields of 2.7 t/ha in 2015 and 2.9 t/ha in 2016, compared to average boro yields of 6.3 t/ha
255
and 3.8 t/ha, respectively. Lower overall yields mean that the same proportional change in yield
256
will be smaller and thus harder to detect. Despite the lack of correlation between aman yield
257
differences and soil As differences, we did observe an increase in aman yields in the high-
258
replaced-by-low plots and a decrease in aman yields in the low-replaced-by-high plots compared
259
to their control plots in first aman season after the soil was exchanged (Figure 3).
260 261 262
Different duration of impact of exchange on soil As and rice yield We observe a strong relationship between boro 2015 soil As differences and yield
263
differences in boro 2015 and boro 2016. Surprisingly, the effect of the soil exchange on yield
264
persisted into 2016 even after soil As concentrations had increased in the high-arsenic plots and
265
the effect of the exchange on soil As was no longer observable (Figure 3).
266
This suggests that some factor other than bulk soil As may mediate the effect of the soil
267
exchange on arsenic available to the rice plants and thus on rice yield. One possible factor is the
268
interplay between soil As and iron oxides, since the presence of As can affect iron oxide
ACS Paragon Plus Environment
Page 12 of 27
Page 13 of 27
Environmental Science & Technology
269
formation and transformation26–28. In general, lowering the concentration of As in a system is
270
expected to result in faster transformation (from less crystalline to more crystalline) and
271
recrystallization of the iron oxides that form as iron from irrigation water precipitates in our
272
study plots. These transformation and recrystallization processes could allow for As uptake and
273
sequestration by the iron minerals, similar to what has been observed for other elements29. As
274
arsenic is added to the soil again over time, it the rate of iron oxide recrystallization could slow,
275
potentially decreasing the uptake and sequestration of newly added As, but also inhibiting the
276
release of As that has already been incorporated. Overall, this could result in a lag between the
277
increase in bulk soil As and the increase in plant-available As, thus resulting in a lag in the effect
278
of the increase in bulk soil As on rice yield. Our study was not designed to examine the
279
relationship between iron oxides and As in these systems, but our results suggest that tracking
280
porewater arsenic and investigating the relationship between soil As and iron oxides may be
281
valuable avenues of future research.
282 283
Countrywide impact of As on rice yield
284
The rice-growing regions of Bangladesh are all composed of soils derived from relatively
285
recent sediments delivered by rivers, and the variability in these sediments occurs on the scale of
286
hundreds of meters30. Our sites span tens of kilometers and contain soils of a range of grain sizes,
287
and yet we have observed a consistent relationship between yield difference and soil arsenic
288
difference across these sites, suggesting that this relationship is generalizable across a variety soil
289
types. Since BR 28 and BR 29 make up more than half of the total boro rice cropped area in
290
Bangladesh24, we can use the relationship we observe between soil As and yield in our study to
291
estimate the overall boro rice yield loss in Bangladesh due to the buildup of irrigation water As.
ACS Paragon Plus Environment
Environmental Science & Technology
292
Wells shallower than 100 m in depth have an average As content of 61 µg/L31. Assuming that
293
paddy soil has a bulk density of 1 kg/L, that boro rice has been irrigated with 1 m of water
294
annually for 25 years, and that all As is retained in the top 15 cm of soil, an estimated average of
295
10.2 mg/kg of As has been added to paddy soil in Bangladesh since the Green Revolution when
296
boro rice irrigation became widespread.
297
As an upper bound estimate, we assume no As loss from the soil, since As concentrations
298
in rice plants are low enough to result in negligible removal with the rice harvest8. Furthermore,
299
while measurements at a site flooded to 4.5 m during the monsoon season suggest that it may
300
lose 13-46% of the As deposited each year32,33, only 9% of land in Bangladesh is flooded to more
301
than 1.8 m32, and shallowly flooded areas appear to retain their As21. The 10.2 mg/kg increase in
302
total soil As corresponds to a change in yield of -0.58 t/ha using our boro 2016 slope of -0.1
303
(t/ha)(mg/kg)-1. Since the observed relationship between soil As and yield is linear, this estimate
304
of the impact of soil As on yield holds true regardless of how the total mass of soil As is
305
distributed across the rice-growing areas. Boro rice was grown on 4.8 × 106 ha in 2012-20132
306
and the corresponding loss attributable to the build-up of As is 4.9 × 106 tons, or 26% of total
307
boro yield. As a lower bound estimate, we assume that only 50% rather than all of the soil As is
308
retained and use our lower observed slope, from boro 2015, of -0.057 (t/ha)(mg/kg)-1 resulting in
309
a loss of 1.4 × 106 tons or 7.4% of total boro yield.
310 311 312
Mitigating the impact of As on rice agriculture Our study conducted in multiple fields across a 150 km2 area shows that soil As
313
negatively impacts boro rice yield. Given that the buildup of irrigation water As in soils may
314
already have substantially reduced rice yield and that the trend is set to continue unless farmers
ACS Paragon Plus Environment
Page 14 of 27
Page 15 of 27
Environmental Science & Technology
315
find a source of low-As irrigation water, it will be important to continue to explore options to
316
address this problem.
317
In our study area, we have occasionally observed farmers removing the topsoil from their
318
rice fields to build up land for houses and other infrastructure. In highly arsenic contaminated
319
fields, targeted removal of the upper 15 cm of soil where the majority of the As buildup occurs
320
could reduce soil As and improve yields. A potential concern with this approach is that the
321
surface soil generally has the highest nutrient concentrations and best soil structure, and thus its
322
removal might cause a substantial enough yield loss to offset yield gains from decreasing the soil
323
As concentration. Further research could be done to better understand these potential impacts of
324
soil removal.
325
We have also observed farmers in our study area switching away from rice to other crops
326
that require less water and are grown under more oxidizing conditions, and thus are less
327
impacted by As. In places where farmers continue to grow rice, other options for reducing the
328
impacts of As include soil amendments, improved water management or treatment, or growing
329
different rice cultivars that are more resistant to the effects of As3,34,35.
330
Acknowledgments
331
This research was supported by NSF Dynamics of Coupled Natural and Human Systems (CNH)
332
grant ICER 1414131 and NIEHS Superfund Research Program grant P42 ES010349. Brittany L.
333
Huhmann was supported by an NSF Graduate Research Fellowship. This is contribution number
334
xxxx of Lamont-Doherty Earth Observatory.
335 336
Corresponding Author
337
*Phone: 617-258-0392; e-mail:
[email protected] ACS Paragon Plus Environment
Environmental Science & Technology
338
Address: Civil and Environmental Engineering, 48-208, Massachusetts Institute of Technology,
339
Cambridge, MA, 02139, United States
340
Supporting Information
341
Figures showing the locations of the study plots, nutrients in replaced versus control plots, yield
342
as a function of soil As, yield difference as a function of soil As difference, and soil arsenic
343
concentrations at different distances from the center of the study plots. Tables showing irrigation
344
water As for the field sites, soil As for the control plots, elemental XRF measurements on
345
standards compared to the reported values, the plots for which yield was measured during each
346
season, and soil nutrient concentrations in replaced versus control plots.
347
ACS Paragon Plus Environment
Page 16 of 27
Page 17 of 27
Environmental Science & Technology
348
References
349
(1)
350 351
Brammer, H.; Ravenscroft, P. Arsenic in groundwater: A threat to sustainable agriculture in South and South-east Asia. Environ. Int. 2009, 35 (3), 647–654.
(2)
Polya, D. A.; Berg, M.; Gault, A. G.; Takahashi, Y.; Mondal, D. Rice is a major exposure
352
route for arsenic in Chakdaha block, Nadia district, West Bengal, India: A probabilistic
353
risk assessment. Appl. Geochemistry 2008, 23 (11), 2987–2998.
354
(3)
355 356
Duxbury, J. M.; Panaullah, G. Remediation of arsenic for agriculture sustainability, food security and health in Bangladesh; Rome, 2007.
(4)
Heikens, A.; Panaullah, G. M.; Meharg, A. a. Arsenic behaviour from groundwater and
357
soil to crops: impacts on agriculture and food safety. Rev. Environ. Contam. Toxicol.
358
2007, 189, 43–87.
359
(5)
BBS. Estimates of Boro Rice (Husked), 2013-2014; Dhaka.
360
(6)
Food and Agriculture Organization of the United Nations. FAOSTAT
361 362
http://faostat3.fao.org. (7)
Meharg, A. A.; Rahman, M. M. Arsenic Contamination of Bangladesh Paddy Field
363
Soils: Implications for Rice Contribution to Arsenic Consumption. Environ. Sci. Technol.
364
2003, 37 (2), 229–234.
365
(8)
Williams, P. N.; Islam, S.; Islam, R.; Jahiruddin, M.; Adomako, E.; Soliaman, A. R. M.;
366
Rahman, G. K. M. M.; Lu, Y.; Deacon, C.; Zhu, Y.-G.; et al. Arsenic limits trace mineral
367
nutrition (selenium, zinc, and nickel) in Bangladesh rice grain. Environ. Sci. Technol.
368
2009, 43 (21), 8430–8436.
369 370
(9)
Khan, M. A.; Islam, M. R.; Panaullah, G. M.; Duxbury, J. M.; Jahiruddin, M.; Loeppert, R. H. Fate of irrigation-water arsenic in rice soils of Bangladesh. Plant Soil 2009, 322 (1),
ACS Paragon Plus Environment
Environmental Science & Technology
371 372
263–277. (10)
373 374
Abedin, M.; Feldmann, J.; Meharg, A. Uptake kinetics of arsenic species in rice plants. Plant Physiol. 2002, 128, 1120–1128.
(11)
Adomako, E. E.; Solaiman, a. R. M.; Williams, P. N.; Deacon, C.; Rahman, G. K. M. M.;
375
Meharg, A. a. Enhanced transfer of arsenic to grain for Bangladesh grown rice compared
376
to US and EU. Environ. Int. 2009, 35 (3), 476–479.
377
(12)
Hossain, M. B.; Jahiruddin, M.; Panaullah, G. M.; Loeppert, R. H.; Islam, M. R.;
378
Duxbury, J. M. Spatial variability of arsenic concentration in soils and plants, and its
379
relationship with iron, manganese and phosphorus. Environ. Pollut. 2008, 156 (3), 739–
380
744.
381
(13)
Azizur Rahman, M.; Hasegawa, H.; Mahfuzur Rahman, M.; Mazid Miah, M. a.; Tasmin,
382
a. Arsenic accumulation in rice (Oryza sativa L.): Human exposure through food chain.
383
Ecotoxicol. Environ. Saf. 2008, 69 (2), 317–324.
384
(14)
van Geen, A.; Zheng, Y.; Cheng, Z.; He, Y.; Dhar, R. K.; Garnier, J. M.; Rose, J.;
385
Seddique, A.; Hoque, M. a; Ahmed, K. M. Impact of irrigating rice paddies with
386
groundwater containing arsenic in Bangladesh. Sci. Total Environ. 2006, 367 (2–3), 769–
387
777.
388
(15)
Abedin, M. J.; Cresser, M. S.; Meharg, A. a; Feldmann, J.; Cotter-Howells, J. Arsenic
389
accumulation and metabolism in rice (Oryza sativa L.). Environ. Sci. Technol. 2002, 36
390
(5), 962–968.
391
(16)
Rahman, M. A.; Hasegawa, H.; Rahman, M. M.; Rahman, M. A.; Miah, M. a M.
392
Accumulation of arsenic in tissues of rice plant (Oryza sativa L.) and its distribution in
393
fractions of rice grain. Chemosphere 2007, 69, 942–948.
ACS Paragon Plus Environment
Page 18 of 27
Page 19 of 27
394
Environmental Science & Technology
(17)
Delowar, H. K. M.; Yoshida, I.; Harada, M.; Sarkar, A. A.; Miah, M. N. H.; Razzaque, A.
395
H. M.; Uddin, M. I.; Adhana, K.; Perveen, M. F. Growth and uptake of arsenic by rice
396
irrigated with As-contaminated water. J. Food Agric. Environ. 2005, 3 (2), 287–291.
397
(18)
398 399
Islam, M. R.; Islam, S.; Jahiruddin, M.; Islam, M. A. Effects of Irrigation Water Arsenic in the Rice-rice Cropping System. J. Biol. Sci. 2004, 4 (4), 542–546.
(19)
Montenegro, O.; Mejia, L. Contamination of rice (Oryza sativa L) with Cadmium and
400
Arsenic by irrigation with the Bogota River water in rice soils of the Lower Basin. Suelos
401
Ecuatoriales 2001, 35 (37), 26–31.
402
(20)
Iqbal, M.; Rahman, G. M.; Panaullah, G.; Rahman, M. M.; Biswas, J. C. Response of
403
Yield and Yield Attributes of Different Rice Genotypes to Soil Arsenic. 2016, 11 (1), 1–
404
17.
405
(21)
Panaullah, G. M.; Alam, T.; Hossain, M. B.; Loeppert, R. H.; Lauren, J. G.; Meisner, C.
406
A.; Ahmed, Z. U.; Duxbury, J. M. Arsenic toxicity to rice (Oryza sativa L.) in Bangladesh.
407
Plant Soil 2009, 317 (1–2), 31–39.
408
(22)
George, C. M.; Zheng, Y.; Graziano, J. H.; Rasul, S. Bin; Hossain, Z.; Mey, J. L.; van
409
Geen, A. Evaluation of an arsenic test kit for rapid well screening in Bangladesh. Environ.
410
Sci. Technol. 2012, 46 (20), 11213–11219.
411
(23)
Dittmar, J.; Voegelin, A.; Roberts, L.; Hug, S. J.; Saha, G. C.; Ali, M. A.; Badruzzaman,
412
A. B. M.; Kretzschmar, R. Spatial Distribution and Temporal Variability of Arsenic in
413
Irrigated Rice Fields in Bangladesh. 2. Paddy Soil. Env. Sci Technol 2007, 41, 5967–5972.
414
(24)
Hossain, M.; Jaim, W.; Alam, M. S.; Rahman, A. M. Rice Biodiversity in Bangladesh :
415
Adoption, Diffusion and Disappearance of Varieties A Statistical Report from Farm
416
Survey in 2005. 2013, No. February, 118.
ACS Paragon Plus Environment
Environmental Science & Technology
417
(25)
418 419
Kalnicky, D. J.; Singhvi, R. Field portable XRF analysis of environmental samples. 2001, 83, 93–122.
(26)
Masue-Slowey, Y.; Loeppert, R. H. R. H. H.; Fendorf, S. Alteration of ferrihydrite
420
reductive dissolution and transformation by adsorbed As and structural Al: Implications
421
for As retention. Geochim. Cosmochim. Acta 2011, 75 (3), 870–886.
422
(27)
Das, S.; Hendry, M. J.; Essilfie-Dughan, J. Effects of adsorbed arsenate on the rate of
423
transformation of 2-line ferrihydrite at pH 10. Environ. Sci. Technol. 2011, 45 (13), 5557–
424
5563.
425
(28)
Auffan, M.; Rose, J.; Proux, O.; Borschneck, D.; Masion, A.; Chaurand, P.; Hazemann, J.;
426
Chaneac, C.; Jolivet, J.; Wiesner, M. R.; et al. Enhanced Adsorption of Arsenic onto
427
Maghemites Nanoparticles : As(III) as a Probe of the Surface Structure and Heterogeneity.
428
2008, 67 (Iii), 3215–3222.
429
(29)
430 431
Latta, D. E.; Gorski, C. a; Scherer, M. M. Influence of Fe2+-catalysed iron oxide recrystallization on metal cycling. Biochem. Soc. Trans. 2012, 40 (6), 1191–1197.
(30)
Weinman, B.; Goodbred, S. L.; Zheng, Y.; Aziz, Z.; Steckler, M.; van Geen, A.; Singhvi,
432
A. K.; Nagar, Y. C. Contributions of floodplain stratigraphy and evolution to the spatial
433
patterns of groundwater arsenic in Araihazar, Bangladesh. GSA Bull. 2008, 120 (11–12).
434
(31)
BGS; DPHE. Arsenic contamination of groundwater in Bangladesh Vol 2 : Final report. In
435
British Geological Survey Technical Report WC/00/19; Kinniburgh, D. G., Smedley, P.
436
L., Eds.; British Geological Survey: Keyworth, 2001; Vol. 2.
437
(32)
Dittmar, J.; Voegelin, A.; Roberts, L. C.; Hug, S. J.; Saha, G. C.; Ali, M. A.;
438
Badruzzaman, a B. M.; Kretzschmar, R. Arsenic accumulation in a paddy field in
439
Bangladesh: seasonal dynamics and trends over a three-year monitoring period. Environ.
ACS Paragon Plus Environment
Page 20 of 27
Page 21 of 27
Environmental Science & Technology
440 441
Sci. Technol. 2010, 44 (8), 2925–2931. (33)
Roberts, L. C.; Hug, S. J.; Dittmar, J.; Voegelin, A.; Kretzschmar, R.; Wehrli, B.; Cirpka,
442
O. a.; Saha, G. C.; Ashraf Ali, M.; Badruzzaman, a. B. M. Arsenic release from paddy
443
soils during monsoon flooding. Nat. Geosci. 2010, 3 (1), 53–59.
444
(34)
445 446
Brammer, H. Mitigation of arsenic contamination in irrigated paddy soils in South and South-East Asia. Environ. Int. 2009, 35 (6), 856–863.
(35)
Spanu, A.; Daga, L.; Orlandoni, A. M.; Sanna, G. The Role of Irrigation Techniques in
447
Arsenic Bioaccumulation in Rice ( Oryza sativa L.). Environ. Sci. Technol. 2012, 46 (15),
448
8333–8340.
449
(36)
BGS; DPHE. Arsenic contamination of groundwater in Bangladesh Vol 1: Summary. In
450
British Geological Survey Technical Report WC/00/19; Kinniburgh, D. G., Smedley, P.
451
L., Eds.; British Geological Survey: Keyworth, 2001; Vol. 1.
452 453 454 455 456 457 458 459 460 461
ACS Paragon Plus Environment
Environmental Science & Technology
462 463
Figure 1. Soil exchange schematic and distribution of the thirteen study sites within a 150
464
km2 area in Faridpur, Bangladesh. The top 15 cm of soil was exchanged between a 5×5 m
465
high-arsenic plot near the irrigation inlet and a 5×5 m low-arsenic plot far from the irrigation
466
inlet. Adjacent 5×5 m control plots remained undisturbed and were managed identically to the
467
soil replacement plots. Heat map of As in groundwater is reprinted from BGS; DPHE. Arsenic
468
contamination of groundwater in Bangladesh Vol 1: Summary. In British Geological Survey
469
Technical Report WC/00/19; Kinniburgh, D. G., Smedley, P. L., Eds.; British Geological Survey:
470
Keyworth, 2001; Vol. 1.36 Copyright 2001 BGS and DPHE. Map data is from Google,
471
CNES/Airbus, and DigitalGlobe.
472 473
ACS Paragon Plus Environment
Page 22 of 27
Page 23 of 27
Environmental Science & Technology
Depth (cm)
Depth (cm)
474
475
Figure 2. Depth profiles over the top 20 cm from two representative study sites. Arsenic
476
profiles measured over the top 20 cm of soil for the low As control (dark blue), low-replaced-by-
477
high (light blue), high-replaced-by-low (light red), and high As control (dark red) plots during
478
the boro 2016 season for a. West Aliabad and b. Ikri. These figures represent the average across
479
monthly samples taken four times from each plot during the growing season. Error bars represent
480
standard deviation divided by the square root of the number of samples.
481
ACS Paragon Plus Environment
Environmental Science & Technology
a
Harvest As Difference (mg/kg) Replaced - Control
482
483 b
Yield Difference (t/ha) Replaced - Control
484
485 486
Figure 3. Soil As and yield differences between replacement and control plots. a.
487
Differences in soil As between the replaced and adjacent control plots over the top 20 cm as
488
measured by XRF on cores collected at harvest. b. Differences in rice yield between the replaced
489
and adjacent control plots. Data are shown for all plots where yield was measured in each
490
growing season, and the numbers below each box indicate the number of pairs of plots that box
ACS Paragon Plus Environment
Page 24 of 27
Page 25 of 27
Environmental Science & Technology
491
represents. The tops and bottoms of each box are the 25th and 75th percentiles. The line in the
492
middle of the box shows the sample median. Outliers are values that are more than 1.5 times the
493
interquartile range beyond the edge of the box. Asterisks denote a significant difference in
494
medians at p = 0.05 according to an unequal variance t test.
495
ACS Paragon Plus Environment
Environmental Science & Technology
3
3 a
b
R 2 =0.45 n =12
2
1
0
0
-1
-1
-2
-2
-20
0
20
R 2 =0.011 n =13
2
1
-3 -40
40
3
-3 -40
-20
0
20
40
3 c
d
R 2 =0.87 n =7
2
1
0
0
-1
-1
-2
-2
-20
0
20
R 2 =0.0022 n =10
2
1
-3 -40
496
Page 26 of 27
40
Boro 2015 Harvest Soil As Diff (mg/kg) Replaced - Control
-3 -40
-20
0
20
40
Boro 2015 Harvest Soil As Diff (mg/kg)
497
Figure 4. Rice yield difference as a function of soil As difference between replacement and
498
control plots. a. Boro 2015 yield difference correlates with boro 2015 soil As difference (y = -
499
0.06±0.02x + 0.3±0.2, R2 = 0.45, p-value = 0.011, n = 13). b. Aman 2015 yield difference does
500
not correlate with boro 2015 soil As difference (R2 = 0.011, p-value = 0.73, n = 13). c. Boro
501
2016 yield difference correlates with boro 2015 soil As difference (y = -0.10±0.02x + 0.02±0.14,
502
R2 = 0.87, p-value = 0.002, n = 7). d. Aman 2016 yield difference does not correlate with boro
503
2015 soil As difference (R2 = 0.002, p-value = 0.90, n = 10). Data are shown for all study plots
504
for which yield was measured that season and for which soil arsenic was measured in boro 2015.
ACS Paragon Plus Environment
Page 27 of 27
Environmental Science & Technology
505
Red symbols represent the four pairs of plots where soil As and yield were measured in all four
506
seasons. Soil As was measured by XRF on the cores collected at rice harvest. Slopes and
507
intercepts are listed with 95% confidence intervals. Error bars represent standard deviation
508
divided by the square root of the number of samples, and regressions are weighted by the error in
509
soil As.
ACS Paragon Plus Environment